爱可可 AI 前沿推介(10.31)

文章预览

LG - 机器学习 CV - 计算机视觉 CL - 计算与语言 AS - 音频与语音 RO - 机器人 1、[LG] Cross-Entropy Is All You Need To Invert the Data Generating Process 2、[CL] Distinguishing Ignorance from Error in LLM Hallucinations 3、[LG] Flow-DPO：Improving LLM Mathematical Reasoning through Online Multi-Agent Learning 4、[CL] Learning and Unlearning of Fabricated Knowledge in Language Models 5、[LG] On the Role of Depth and Looping for In-Context Learning with Task Diversity 摘要：反演数据生成过程有交叉熵就够了、大语言模型中无知型幻觉与错误型幻觉的区别、通过在线多智能体学习改进大语言模型数学推理、语言模型中虚构知识的学习与遗忘、深度和循环在任务多样性上下文学习中的作用 1、[LG] Cross-Entropy Is All You Need To Invert the Data Generating Process P Reizinger, A Bizeul, A Juhos, J E. Vogt… [Max Planck Institute for Intelligent Systems & ETH Zurich] 反演数据生成过程 ………………………………